Seattle Times and Newsday Sue OpenAI and Microsoft

If the US government has just told a court that training on news is fair use, why would two regional newspapers sue over exactly that a few days later?

That is the question raised by the complaint The Seattle Times and Newsday filed against OpenAI and Microsoft on September 5 in the US District Court for the Southern District of New York. The claim is copyright infringement: years of reporting used to train AI models without permission or payment. The complaint describes generative AI as consuming "human-authored content" and returning "copies and derivative imitations," which points at both the training data and the outputs.

What the filing says, and what it leaves open

The venue is the same court hearing The New York Times' 2023 case against the same two defendants. Microsoft told TechCrunch it was "surprised by the lawsuit" but is "always happy to sit down and explore solutions." OpenAI has not commented.

Several details are not yet public: the damages sought, the law firm, and whether the case will be consolidated with the existing New York Times litigation. The Southern District venue is consistent with either path, so "they joined the NYT case" would be a guess. What is documented is a separate complaint in the same courthouse.

The timing is the point

Earlier this week the Department of Justice filed a statement of interest in the NYT case supporting OpenAI's fair-use position. A government brief like that could have discouraged smaller publishers. The opposite happened: two regional papers went to court within days. Read one way, the publishers are treating the DOJ position as one party's view rather than a signal about how the court will rule. Read another way, they are moving before any ruling narrows their options.

Either reading tells you something about the market. The list of plaintiffs now runs from national outlets to regional dailies. Publishers who are not suing are, in many cases, signing licensing deals instead. The middle ground of doing nothing is shrinking.

Why a business outside publishing should care

Two consequences reach beyond newsrooms. First, every new suit nudges model providers toward paid licensing, and licensing costs eventually show up in API and subscription prices. Second, the argument that model outputs are "derivative imitations" is a claim about what comes out of the model, not only what went in. Companies publishing AI-drafted content at volume should expect that question to be tested in court, and a provenance check before publishing is a cheap habit to build now.

Sources: TechCrunch